or at least more common systems, in part because their eVects much more
contingent (Bowler 1996 ; Bowler, Donovan, and Brockington 2004 ).
Cox argues that coordination problems lie at the heart of any electoral
system—even SMSP. Some electoral systems tend to raise more problems of
coordination than others and so demand more of both voters and parties.
Outcomes under such systems are therefore much more highly contingent under
some systems than others and are especially chronic under multicandidate
systems that allow voters to choose over candidates. For example, under both
SMSP and also list PR some thought may go into what kinds of candidates to
nominate but little thought has to go into the number of candidates to nominate.
Some disagreement may take place (and hence some coordination be required)
over which candidates to nominate—which local notables or party stalwarts—but
almost none over the number. Under systems such as STV or cumulative voting
(CV) 5 the eventual outcome depends in part on the number of candidates each
party nominates. These systems have multiseat districts (district magnitude> 1 )
and also allow voters to express an intensity of preference over several candidates.
Under STV, voters are allowed to rank order candidates; under CV, voters are
allowed to cast as many votes as there are seats and either give one vote to each
preferred candidate or cumulate those votes on one or two candidates. These
features permit a much greater deal of strategic leeway on the part of both voters
and candidates and so outcomes under these systems are contingent on the abilities
of the players to play the game as well as upon the rules themselves. Under STV and
CV, for example, parties can do either better or worse than a purely proportional
outcome depending on their ability to strategize and be disciplined (Bowler 1996 ;
Bowler, Donovan, and Brockington 2004 ).
Some systems, then, would seem to produce outcomes that are less dependent
on how players play the game than others: perhaps because, following Cox’s
interpretation, they simply require fewer coordination problems to be solved.
The fact that commonly used electoral systems such as list PR and SMSP produce
clear outcomes in addition to our well-developed understanding of proportionality
(Gallagher 1991 ; Lijphart 1985 ; Blais 1988 ; Farrell 2001 ) can lead to a false sense of
conWdence in our ability to engage in electoral engineering. While we can say small
or large parties will beneWt under various regimes, we cannot predict with any
precision the question of interest to most politicians: which large party and which
small party?
Uncertainty about electoral system eVects seemed especially prevalent in ‘‘big
bang’’ changes where democracy is introduced. After a while this uncertainty—at
least theWrst two forms of uncertainty—may well be reduced but electoral systems
nevertheless seem to remain relatively stable, despite the seemingly ever-present
incentive to jockey for or shore up an advantage, through the electoral process. The
5 Used in local elections in the modern USA and Victorian Britain and many corporate settings
(Bowler, Donovan, and Brockington 2004 ).
electoral systems 585